How far can 5,500 hours of real-world driving data take you?
All the way to VATIX, a new open-source, state-of-the-art world model for driving video generation. 🚗
The first major result from the NATIX × @Valeo_Group collaboration is here!
There is no lack of data in the world. There is a lack of open data collection methods.
NATIX solves that problem, so Physical AI innovation can move forward. ⏩
That's the promise of Decentralized Physical AI 💪
5/
The first release includes 5 safety-critical events across 10 U.S. states.
Just like our previous datasets, we’re making it openly available for researchers & open-source teams.
Read more about what makes this dataset unique 👇
https://t.co/QXBwo2MTuB
Our 3rd open-source driving dataset is live on @huggingface! 🚗
This time, we’re adding a new layer to NATIX data: vehicle telemetry.
Real-world multi-camera footage, now paired with precise signals showing what the vehicle was actually doing.
4/
Telemetry and VLMs are complementary.
Telemetry can flag the exact moment a harsh brake happens. A VLM can then analyze those few seconds of footage to explain why.
Vehicle telemetry finds the event. Visual AI explains it. 🤝
World models are more than a hype word. 📈
The biggest companies are investing in a technology that can simulate the future for Physical AI.
And beneath it all sits real-world data that keeps World Models grounded in reality.
From driving footage to simulated scenarios, that's the promise of World Models.
This allows autonomous vehicles to commit to any action, and the scene evolves as it would in the real world. 🌐
That is exactly what VATIX 9B, a world model built by Valeo using NATIX data, does.
World Models can take one scene and create endless possible futures.
That is extremely valuable for the training, testing, and validation of autonomous driving systems.
Data goes in. Intelligence comes out.
Hand-built simulations teach a vehicle how to drive according to pre-written laws.
World Models teach autonomous driving how to analyze, anticipate, and react.
This is how autonomy scales 📈
August Progress Update is here:
📹>196K Hours of Multi-Camera Footage
💹>8.4B $NATIX Staked
🤖How autonomous driving data helps robotics
🔄The difference between a dataset and a data engine
Full recap👇
https://t.co/TIHcHXPAOl
NEW: @NATIXNetwork says VATIX, its new open-source world model for driving video generation, is the first major result of its collaboration with Valeo. The project draws on 5,500 hours of real-world driving data.
https://t.co/XnHPA6mX9Y
9/
5,500 hours of community-collected driving data helped push open-source driving video generation to new heights.
This is what Decentralized Physical AI can make possible. 💪
Explore VATIX and the results 👇
https://t.co/eMABLMGjvr
How far can 5,500 hours of real-world driving data take you?
All the way to VATIX, a new open-source, state-of-the-art world model for driving video generation. 🚗
The first major result from the NATIX × @Valeo_Group collaboration is here!
8/
VATIX 9B is the largest open-source video diffusion model trained from scratch on driving data.
It sets a new open-source benchmark for driving video generation. ⚡️
This is the first result from our broader work with Valeo, with a wider multi-camera WFM still ahead.